Better cluster manager, Pre-instantiation fix, and better monitoring

[jimfcarroll]

This is now ready and contains these major changes:

1. Refactor of the Cluster manager
2. (1) results in the need to refactor the Routing Strategy
3. Fix for the race condition between the Routing Strategy and Pre-instantiation in issue Possible race conditions during pre-instantiation (micro-shard allocation and mp instantiation) #60
4. Added sent/recieved byte count metrics. Closes issue Add messages byte count to the stats #63.
5. Separately track mp failures and dempsy failures during message processing. Closes issue Account for Message Failed due to Framework or message processor #62.
6. Add the ability to measure the mp message handling time. This closes issue Collect stats on MP latency #61
7. The monitoring for Adaptors in a distributed application should have a prefix that's consistent with the monitors for non-Adaptors in the same application. This closes issue adaptor showing up under graphite - under local folder #66.